FR2807849A1 - METHOD AND SYSTEM FOR SEARCHING AND HELPING CHOICE - Google Patents

Abstract

The invention concerns a system comprising means for encoding data concerning said objects by (a) setting up a table (T) with hierarchical structure of the set of attributes (Cna) and contents (Cnu) associated with said objects; and (b) setting up for each stored object a file wherein each attribute of said object and its contents is identified by the specific code assigned to said attribute-contents pair in said table. Said system comprises calculating and interfacing means for, in response to a search request from an applicant, (c) constructing a first table containing the set of stored files concerning a request; (d) constructing a second table containing the set of contents present in any one of said files of the first table; (e) presenting the successive selections of attributes and contents to said applicant, in the hierarchical order defined in said second table; (f) updating the first table by selecting in said first table the files comprising said selected attribute-contents pair; (g) repeating said operations e) and f) until at least one predetermined condition is fulfilled; and (h) presenting the result of said search to said applicant when said condition is fulfilled.

Description

The present invention relates to a data processing method for searching and assisting the selection of objects from among a set of objects stored as cards in a data processing system, in which each object is associated with the contents of each of several attributes, as well as a data processing system and human / machine interface for the implementation of this method.

In the following, the term "object" designates any material or immaterial entity that can be described by a certain number of attributes, hereinafter also called containers, the attributes possibly having one of several values given hereinafter called contents. In an ordinary database, all the attributes or containers are common to all the cards and the information of a card consists of the list of associated contents. For example, searching on these databases involves using keywords, crossing criteria, and so on. In many cases, searches on these conventional databases may lead to no response or too many responses.

Thus, to find a website on the Internet, it is necessary to consult search engines that, depending on the case, give no response or offer several thousand or tens of thousands of responses. The user, in the presence of a long list of answer, can not refine his question and, moreover, he does not know what relevance to wait for the proposed answers.

The aim of the invention is to provide a method of data processing for the search of objects that guarantees to always present to the users answers, and this in the form of short lists of relevant answers, while using computer means of limited power and offering a great speed of treatment and a reasonable quality / cost ratio.

For this purpose, the subject of the invention is a method for processing data for selecting objects from among a set of objects stored in card form in a data processing system, in which each object associated with the content of each of them several attributes, <U> characterized in that </ U> A) the data relating to said objects is coded by a) establishing a hierarchical structure of the set of attributes and contents associated with said objects, in which said attributes are classified hierarchically in several ranks, each attribute of a secondary rank is attached to a hierarchically superior rank attribute content, and each pair constitutes an attribute and its content is associated with a specific code, b) establishing for each memorized object a sheet in which each attribute of said object and its content identified by the specific code assigned to said pair attribute-content in said table, and B) in response to an inquiry request from an applicant: c) constructing a first table containing all the stored records relating to said request, d) constructing, by means of said codes, in accordance with said hierarchical structure of pairs attribute-content second table (C) of the contents present in any one of said sheets of the first table (S), e) presents successive choices of attributes and contents to said requestor, in the hierarchical order defined in said second table (C), f) updating said first table in response to the choice by said requestor of an attribute-content pair consecutive to said step e), by selecting in said first table (S) records comprising said attribute-pair selected content, g) repeating said steps e) and f) until at least one predetermined condition is satisfied, and h) presenting the result of said search to said applicant. t when said condition is satisfied.

Thus, according to the method of the invention the search is carried out in stages, each step involving the choice of an attribute or containing among those available and the choice of a content among those available in the chosen attribute or container. Each step leads to presenting only relevant information to the applicant since it is presented only search criteria that lead to responses and that search criteria are presented to him only as long as there are answers.

According to one characteristic of the invention, said step e) comprises the step of e1) establishing, by exploration of said second table at the highest hierarchical rank not yet explored, a list of attributes having at least two contents represented in said first table (S), and one of two steps of e2) presenting said list of attributes to said requestor for selecting one of said attributes if said attribute list comprises at least two attributes, and e3) directly presenting the contents of an attribute to said requestor for the purpose of selecting one of said contents if said list of attributes comprises only said attribute.

According to one embodiment of the invention, said step e2) comprises the steps of e21) establishing, in response to the choice of one of said attributes, said requestor, a list of contents of said chosen attribute represented in said first table (S ), e22) presenting the said list of contents to the applicant for the choice of one of the said contents.

According to another embodiment of the invention - said step e1) comprises the establishment, for each of said attributes of said list of attributes, of a list of the contents of this attribute which are represented in said first table (S) , and - said step e2) comprises the presentation to said applicant of all attribute-content pairs selected in step e1), for the purpose of selecting one of said couples by said applicant.

The invention also includes one or more of the following features considered alone or in combination - said condition of step g) is satisfied if said list of attributes of step e1) is empty; said condition of step g) is satisfied if the number of index cards of said first table is less than a predetermined value - following said steps c) and f), said number of said cards of the first table is presented to said requestor; - following said steps c) and f), the applicant is presented with the list of attribute-content pairs previously selected; - following said steps c) and f), the applicant is presented with the list of attribute-content pairs which, although never selected by the applicant, are nonetheless common to all the remaining cards; defining at least one of said attribute-content pairs constituting a priority search criterion, constructing said first table and said second table of said steps c) and d) by selecting, from said set of cards, a subset of said cards satisfying to said criterion, and performing said steps e) through said subset of cards, - said definition of a priority search criterion includes the step of offering said applicant to choose at least one priority criterion subsequent to said step e), to add the said priority criterion chosen by the said applicant to a list of priority criteria, and to develop the said subset of cards by selecting the cards of the said set satisfying the priority criteria of the said list.

The invention also relates to a data processing system for implementing the method defined above, this system comprising data storage means for storing said table, said sheets and said tables, calculating means for the implementation of said steps c) to g), and interface means for the presentation of said choices and results to said applicant and for the implementation of said choices by said applicant. According to a feature of the system according to the invention, it presents a client-server architecture, said interface means comprising a client terminal.

Other features and advantages of the invention will emerge from the following description of embodiments given solely by way of example and illustrated by the appended drawings in which - FIG. 1 is a table illustrating the hierarchical structure of the set of attributes or containers and contents associated with the objects in a system according to the invention; FIG. 2 is a table similar to that of FIG. 1 illustrating the containers and contents in the case of a practical application; FIGS. 3 to 9 are flowcharts illustrating the method according to the invention, and FIGS. 10 to 13 are tables illustrating an exemplary encoding contained for the implementation of the method according to the invention.

Reference will first be made to table T of FIG. 1, which illustrates the hierarchical structure of all the containers and associated contents objects stored in a computing resource and from which a search is likely to be made.

These objects are represented by cards F constituted - on the one hand, of various information on which it is not possible to carry out research using the mechanisms of the invention, for example text, images, digital data and / or other information of any kind; - on the other hand, contained-content pairs (cna-cnu) likely to have each one of several contents, which belong to the hierarchical table of Figure 1 and which serve as well various information as criteria of research.

The cna-cnu pairs of a form F constitute a subset of the table T of FIG. 1 and have the following property: in the table T, there exist sub-pairs cna-cnu of a specific -cnu pair belonging to to the form, then there exists in the form one of these sub-couples, that is to say that the form is complete. As shown in FIG. 1, in the hierarchically structured table T, the containers and contents are classified hierarchically in several ranks A, B, C ... and each container of a secondary rank is attached to a content of a containing a hierarchically superior rank. Thus, a level or rank A of the table is provided a Container 1 and Container 2 corresponding to the main topics of classification. Container 1 can have several possible values or contents, namely cnu1a, cnu1b, cnu1c, ... At level or hierarchical row B corresponding to sub-themes, there are four containers. The containers cnu1 a1 and cnu1a2 are attached to the content cnu1a. Similarly, the containers cnu1b1 and cnu1b2 are attached to the content cnu1b. At the same level or hierarchical rank B, the container cnu1 a1 may have two possible values or contents, namely cnu1 a1 a or cnu1 a1 b. By traversing the hierarchical table tree T of FIG. 1, it is thus possible to describe the total content-containing pairs of any of the cards F, and thus hierarchical links of these different pairs in the card.

Figure 2 illustrates the application of Table T of Figure 1 to the real estate field. Thus the containers of the level or higher rank A include the type of real estate, the nature of the transaction on this property and the place where the property is located. For container "Type of good", content can be "Apartment" or "House", for the container "Transaction", the content can be "Sale" or "Location", etc. At the next hierarchical level B, the containers (number of pieces, floor, plot, price, etc.) are each attached to one of the contents of the higher hierarchical rank: the "number of pieces" and "floor" containers are attached. to the content "Apartment" and the container "Land" is attached to the content "House".

It follows from the foregoing that in the hierarchical table T of FIG. 1, each pair containing content is unique and can be associated with a specific code making it possible to designate said pairs in the cards F.

Reference will also be made to the flowcharts of FIGS. 3 to 9 which illustrate the mechanisms used to search among a set of F cards coded according to the hierarchical table structure of FIG.

The general flowchart of Figure 3 entitled Successive Selections for Purpose presents the successive choices of containers and contents to an applicant until a short list is obtained. Step 1 corresponds to the start of a search, that is to say to the preparation of tables S and intended to reflect all existing records F from which research must be performed.

The table S is a table of the selected cards F and their couples containing content, this first table being constantly updated during the search.

The second table C is a table of the contents present in any of the cards of the first table S. If each card F is considered as a subset of contents, then the table C can be considered as the set constituted by the union of the subsets F, it being understood that each pair containing-content is represented only once in the table C. Moreover, this table is organized in accordance with the hierarchical structure of the table T to reflect the hierarchy container-content couples. This means that the containers are represented only implicitly in this table, because each content has a specific code, from which can be deduced its container as shown in Table T.

If, following the start of the search in step 1, it turns out that the total number of records stored among which the search performed is less than a predetermined value, the test 2 leads directly in 3 to the presentation to the applicant from the list of all existing cards.

If the number of sheets of the table S elaborated in step 1 is greater than said predetermined value, that is to say, if it is not a short list, test 2 is passed. Step 4 of choice containing / content. This step corresponds to a module of which a first embodiment is illustrated in FIG. 4 and a second embodiment in FIG.

Following the step or module 4, it is examined in 5 if there is still a possibility of choice containing / content. If there is still at least choice test 5 returns to the input of the test 2. Otherwise, we go directly to 3 to the presentation of the list of records found.

If the number of cards presented to the applicant is greater than a predetermined parameter, it means that it is desirable to improve search by refining the criteria containing (attribute) -content dependent those chosen. In other words, it is necessary to add search criteria (containers and their contents) and / or add contents for one or more existing containers (attributes). Optionally, it can be in a planned order to warn the person in charge of the definition of the criteria.

Reference will now be made to FIG. 4 illustrating the first embodiment 4.1 of module 4 of FIG. 3. This module 4.1 begins with a step 10 entitled "Divisible containers" and illustrated by the flow charts of FIGS. 6 and 7. This module 10 is intended to create a container list by hierarchical search in the second table C of containers having at least two contents represented in the first table S.

Referring to Figure 6, the module 10 begins in step 100 by exploring the second highest ranked C table in the hierarchical rank that is not yet explored to determine which contain-content pairs of that rank. At step 101, each container is processed as will be described with reference to Figure 7 and in 102 a list of containers is provided for presentation to the applicant as will be described later.

Referring to Fig. 7, the step or module 101 begins with step 200 of searching for whether the processed container has content in table S. If no content is found at 201, that is ie if result is R0, we go directly to 202 at the end of the concerned containing process. If the response to the test 201 is positive, proceed to step 203 of searching whether the processed container has a second content shown in the table S. If the result of the test 204 is negative, i.e. the container has only one content R1, the container is ignored, but the search is extended to 205 by also processing all containers depending on the content found. the response to test 204 is positive, the result is that found in table S at least two different contents for the treated container. The treated container is added to the results list at 206, and the end of treatment of this container is then reached at 202.

The process 101 for processing a container of the figure is repeated for each of the containers selected in step 100 of FIG. 6, until all the selected containers are exhausted. We then have at 102 the list of all the containers having at least two contents represented in the table S, it being understood that they do not contain the containers having at least two contents, but depending hierarchically on another container which also appears in the list.

Going back to FIG. 4, if it turns out in test 11, the list of containers 102 is empty, that is to say that there is no choice to propose to the applicant as indicated in 12. then in 3 (figure 3) to the presentation of the list of found records.

If under test 11, it turns out that the list of containers 102 is not empty, it is examined in 13 whether this list comprises one or more containers. If there is more than one container in the list, the applicant is shown to 14 and the applicant selects one of the containers.

When the applicant has chosen a container at 15 if the test 13 determines that the list of containers 102 contains only container, we move to the content selection module of step 16 of creating a list of contents represented in the table S for the single container or the container chosen by the applicant.

Referring to Fig. 8, the first step 300 of Content Selection Module 16 is to locate in the selected container table C in the unique container from list 102, hereinafter referred to as CNA, and then to find the first CNU content of this container. The next step 301 is to find out if the CNA-CNU pair is part of a table of the table If it turns out in 302 that this is the case, then go to step 303 where adds this content CNU to the list of contents of the CNA container. Optionally, the number of records in the table including the CNA-CNU pair is also calculated at 303 and a list of records of the table that includes the CNA-CNU pair is created.

then goes to test 304 where it is determined whether the CNA container contains other contents. If so, go to 305 to the next content and then return to step 301 to determine if the new pair CNA- is part of one of the cards of table S. Then repeat steps 303 and 304 until all contents of the CNA container have been examined. If, during this process, it turns out that a CNA- pair is not part of one of the cards in the S-table, the answer to the test 302 is negative and then passes to the test 304, which means that the content in question is not added to the list of results.

Once all the contents of a CNA container have been examined, the list of selected contents is provided at 306 and is presented to the applicant at step 17 in Figure 4. The applicant proceeds to select one of 18 those which are presented to him and one then proceeds to the step 19 of reduction which consists in deleting from the table S all the cards do not contain one of the couples containing content indicated by the applicant. Optionally, it is possible to present to the applicant the list of contained-content pairs which, although never selected by the applicant, are nevertheless common to all the remaining cards. These are couples that have been inferred during the search process to eliminate containers with only possible content; this elimination follows from the choice R1 of step 204 of FIG. 7.

After step 20, which indicates that the applicant has made a choice of container at 15 and content at 18, return to the input of test 2 in FIG. 3 to determine if the list of cards contained in the table S a short list or not. If yes, the search is complete and we go to step 3 of presentation of the found records, while in the negative, we repeat the process of module 4 described above by successively exploring the containers and contents in the hierarchical order until a short list is obtained. A short list is a list whose number of cards is less than or equal to a predetermined value, for example ten cards. FIG. 5 illustrates a second embodiment 4.2 of the module 4 of FIG. 3 in which all the containers divisible by the module 10 of FIG. 6 are determined and all the possible contents of each of the divisible containers found by means of FIG. 16 of Figure 8. As in the embodiment of Figure 4, we go directly to Step 12 back to Step 3 if it turns out 11 that there is no container having at minus two contents represented in the table S. In the opposite case, the content selection module 16 is unrolled for each container found and the applicant is then presented to 21 any container-content pair, that is to say the pairs offering a choice.

In 22, the applicant selects a pair and at 23, the reduction described in step 19 of FIG. 4 is carried out. Step 24 indicates that the choice of container-content pair has been made and, as for the module of Figure 4, it then returns to the input of the test 2. As indicated above, the process of successive selections then continues until a short list is obtained.

The flow diagram of FIG. 9 illustrates an alternative embodiment of the method of FIGS. 3 to 8 according to which the successive selection process can be initialized by application of priority criteria necessarily leading to one or more responses.

These priority selection criteria (pairs containing / content) can be predefined, for example - by means of an extract of previous selections as described below with regard to FIG. 9; - depending on a dependence on the means used to enter the search system; - depending on the profile of the applicant; - according to the geographical location of the applicant, which he has previously specified or which is detected using a GPS or other type of system; - by means of a free preliminary inquiry of the "full text" type on all the contained files containing all the files; - in response to a specification of the applicant only treat records provided by one or more specific bodies, this list is not exhaustive.

Referring to FIG. 9, step 401 corresponds to this optional preset of one or more priority criteria. In the following, we will call P the list of priority criteria.

The next step 402 consists in automatically applying to the set F cards all the priority criteria that have been predefined, in order to prepare tables S and C described above.

With leaves remaining at the end of step 402, the "successive selections" algorithm of FIG. 3 is executed at 403 (steps 401 and 402 corresponding to phase 1 of initialization of the search), which involves successive selection of the applicant as described above.

 After the algorithm of Figure 3, the applicant has the option to choose one or more priority criteria.

If the applicant chooses in 404 to apply one or more priority criteria to the search, this means that he or she has found that these criteria were more important than those he may have selected by the algorithm of FIG. The applicant therefore wishes to apply this or these priority criteria, already obtained, to all the cards F.

A positive response to the 404 test leads to completing in 405 the list P priority criteria already obtained by those who have just been selected. By construction, the list P of priority criteria gives at least one answer. We then return to step 402, that is to say, we automatically apply to all the cards F the priority criteria of the list P completed in 405 and we begin the process of "successive selections" until a negative answer is given to the 404 test. The result is 406 at the end of the search.

Reference will now be made to the tables of FIGS. 10 to 13 which illustrate an example coding of the contents by bit tables for the implementation of the method described above.

FIG. 10 is a representation in the form of a tree structure equivalent to the table T of FIG. 1. In this exemplary embodiment, each content of the hierarchical tree array is numbered in the order of encounter according to a recursive scanning of the tree structure: Since the first content that depends on the hierarchical rank level container Cna1 has the number 1 (Cnu1 in Figure 10), the contents of the Cna1.1.1 container (rank B) depending on Cnu1 are numbered 2 and 3 (Cnu2 and Cnu3), etc. In the example of Figure 10, there are nine contents.

As shown in FIG. 11, the form F describing each object consists of a table having as many memory locations as there are contents in the table or tree of FIG. 10, in this case 9 in the example under consideration. Each of these memory locations can take the logical value .1 if the corresponding content of the table T is included in the form and 0 otherwise. The position of each memory location in the form F, that is to say its address, corresponds to a content number assigned in the table T as shown in FIG. 11. Thus, each memory location of the form F is associated with in a one-to-one way to a container-content pair of array T.

At every moment of the search process, the table S consists of all the cards F that have been selected so far. Referring to FIG. 12, it will be assumed that after a first choice of a container Cna1 by an applicant, there remain four cards F1, F2, F3 and in the table S. cards have the following contents - for: Cnu1, Cnu3, Cnu7 and Cnu8; for: Cnu4, Cnu5, Cnu7 and Cnu9; for: Cnu4, Cnu6, Cnu7 and Cnu9; and for: Cnu1, Cnu2, Cnu7 and Cnu8.

 Since in the previous phase the applicant chose the container Cna1, he is now asked to choose between the contents Cnu1 and Cnu4. Assuming that the applicant chooses Cnu1, then the table S will no longer consist of only sheets F1 and F4, and, as shown in FIG. 13, the table C will be directly deduced from the table S by performing on each of the bits of the same order in each of the sheets selected (F1 and F4) an inclusive OR operation. The table C is thus organized in the same way as a form F: the bits or memory locations constituting this table are arranged in the order of a hierarchical exploration of the contents of the table T. The bit order number also allows find in one or more unrepresented side files (for example one file per language) the description of the content and vice versa. This arrangement also makes it possible to group the bits of the contents that depend on the same container whatever the hierarchical difference between the container and the contents considered. In the example, the order is: Cnu1, Cnu2, Cnu3, Cnu4, Cnu5, Cnu6, Cnu7, Cnu8, Cnu9.

This coding mode has the advantage of allowing a very compact representation of the table T, the sheets F and the tables S and C if the average number of contents per container of the table T is low. One can indeed represent several contents in a single word machine. This mode of organization coding also offers the possibility of making logic on any machine word, thus on several contents at a time, a considerable saving of time to find the contents represented in any one of the remaining sheets. Furthermore, during the implementation of the content selection module of Figure 8, it will be possible to simultaneously test several contents (as much as bits in a word machine) and this module will be accelerated. The disadvantage is that the initialization of the search requires the scanning of all the files, which can be bypassed by constituting pre-sorted auxiliary tables to analyze the first criterion introduced.

In practice, the table S of FIG. 12 consists of a table of pointers to each of the cards F selected from the set of cards from which the search is made.

Initially, the table S contains the list of all the cards F and a marker indicates the total number of cards selected. At each step of the search, all records up to the marker are analyzed, and retained records are pushed to one end of the table and the marker is moved to indicate the new number of records remaining. Optionally, the rejected cards can be packed to the other end of the table, which allows to segment it according to the search steps and thus quickly back on a search step.

It follows from the foregoing that the described method is particularly suitable for multi-criteria search when the search theme allows prioritization of these criteria. Of course, the tree structure or hierarchical structure of the table T varies according to the theme considered, that is to say, it is not the same depending on that topic is example automobile, real estate, l appliances, etc. The described search method may be preceded by a search in traditional mode, example by keyword, to select the theme that will be the root of all containers among which the search will be conducted as described with reference to Figures 3 to 9.

The implementation of this method requires a data processing system comprising data storage means for storing the table T, F sheets and tables C and S, calculation means for implementing the described process. with reference to FIGS. 3 to 9 and interface means for presenting choices to the applicant, to enable him to make these choices and to present him the result of his choices. The data processing system can be of any type and it can be in particular a client-server architecture system used via a network such as the Internet from a client terminal such as a personal computer PC, a mobile phone operating by input or by voice recognition, etc ...

The choices offered to an applicant can be proposed to him on the screen of his terminal in the form of dynamic pages composed for example of one or more of the following windows - a window presenting a reactive geographical map with a shades of colors or pictograms functions the number of responses specific to each location (country, region, department, city); a window representing reactive vertical histograms illustrating prices, measures or qualities (weight, size, area, area, distance, time, period, color, ...); - window of remaining possibilities with the selection criteria (container attributes) still possible and, for each of these criteria, the list of available values or contents with the number of responses, presented in the form of horizontal histograms or drop-down menus - window representing the choices already made by the applicant in the form of a virtual form composed of a list of containers and contents increases as and when successive choices; a window of results synthesized in the form of a two-dimensional table with reactive elements (photos, texts, titles, logos, drop-down menus, etc.) or in the form of dynamic cumulations presented graphically (pie charts, histograms, etc.) ..); - etc .....

The elements of these different windows are reactive in that they make it possible to make a choice by means of a mouse by moving a pointer these elements, as is now conventional.

Of course, the described embodiments are only examples and they could be modified, in particular by substitution of technical equivalents, without departing from the scope of the invention.

Claims (2)

A data processing system for selecting objects from a set of objects stored in the form of cards in said system, wherein each object is identified by the contents of each of a plurality of attributes, <U> characterized in that it comprises </ U> a) means for storing an array with a hierarchical structure of all the attributes and contents associated with said objects, wherein said attributes are hierarchically classified in several ranks, each attribute of a secondary rank is attached to an hierarchically superior rank attribute content, and each pair of attribute and content is associated with a specific code, b) means for storing, for each memorized object, a card in which each attribute of said object and its content is identified by the specific code assigned to said attribute-content pair in said array, and B) computing and interface means for, in response to a req. search of an applicant: construct a first table (S) containing all the stored records relating to said request, d) construct, by means of said codes, in accordance with said hierarchical structure of attribute-content pairs, a second table (C) contents present in any one of said sheets of the first table (S), e) presenting successive choices of attributes and contents to said requestor, in the hierarchical order defined in said second table (C), f refreshing said first table (S) in response to said choice by said requestor of an attribute-content pair consecutively said step e), by selecting in said first table (S) the records comprising said selected attribute-content pair, g) repeating said operations e) and f) to the satisfaction of at least one predetermined condition, and h) presenting the result of the said search to the applicant when the said condition is satisfied. sfaite. 2. System according to claim 1, characterized in that said computing and interface means e) for presenting successive choices of content attributes to said requestor, in the hierarchical order defined in said second table (C), comprise e1. ) means for establishing, by exploring said second table (C) at the highest hierarchical rank not yet explored, a list of attributes having at least two contents represented in said first table (S), and one of the two means comprising e2) means for presenting said list of attributes to said requestor for the purpose of selecting one of said attributes if said list of attributes comprises at least two attributes, and means for directly presenting an attribute to said applicant in view of the choice of said contents if said list of attributes includes only said attribute. System according to claim 2, characterized in that said means for presenting said list of attributes to said requestor for the purpose of selecting said attributes if said list of attributes comprises at least two attributes comprises e21) means for establishing, in response to the choice one of said attributes by said requestor, a list of contents of said selected attribute represented in said first table; e22) means for presenting said list contained to said requestor for the purpose of selecting one of said contents. 4. System according to claim 2, characterized in that - said means e1) comprise means for establishing, for each of said attributes of said list of attributes, a list of the contents of this attribute which are represented in said first table. (S), and - said means e2) include means for presentation to said applicant of all attribute-content pairs selected by said means e1), for the purpose of selecting one of said couples by said applicant. 5. System according to any one of claims 2 to 4, characterized in that said condition of the operation g) is satisfied if said list of attributes established by said means e1) empty. 6. System according to any one of claims 1 to 5, characterized in that said condition of the operation implemented by said means g) is satisfied if the number of records of said first table (S) is less than a value. predetermined. 7. System according to any one of claims 1 to 6, characterized in that said interface calculating means comprise means for, following said operations implemented by said means c) and f), to present to said applicant number of said records of the first table (S). 8. System according to any one of claims 1 to 7, characterized in that said computing and interface means comprise means for, following said operations implemented by said means c) and f), to present to said applicant the list of attribute-content pairs previously selected. 9. System according to any one of claims 1 to 8, characterized in that said interface calculation means comprise means for, following said operations implemented by said means c) and f), to present to said requestor the list of attribute-content pairs which, although never selected by the applicant, are nonetheless common to all the remaining cards (F). 10. System according to any one of claims 1 to 9, characterized in that said interface calculation means comprise means for defining at least one of said attribute-content pairs constituting a priority search criterion, constructing said first table. (S) said second table (C) of said operations implemented said means c) and d) by selecting, from said set of cards, a subset of said cards satisfying said criterion, - performing said operations implemented said means e) to h) using said subset of cards, 11. System according to claim 10, characterized in that said means for defining a priority search criterion comprise means for offering said applicant to choose at least priority criterion consecutively to said operation implemented by said means e), means for adding said priority criterion chosen by said applicant ant to a list (P) priority criteria, and means for developing said subset records by selection of cards said set meeting the priority criteria of said list (P). 12. System according to any one of claims 1 to 11, characterized in that it has a client-server architecture, said interface means comprising a client terminal.